Levi Gans
Nobody goes looking for Datadog alternatives on a quiet day. It usually starts with an invoice that’s bigger than last month’s, and a long afternoon working out which meter moved.
Datadog is a good product. It’s also priced per product, and most products have their own meter. The US price list has more than 100 line items, each with an annual, a month-to-month and an on-demand price.
This guide starts with how Datadog bills, straight from its own docs. Then nine alternatives, grouped by how much of Datadog they replace. Some replace all of it. Some replace only the part you were really using.
Why teams leave Datadog
Start with hosts. Infrastructure Pro costs $15 per host a month on an annual plan and $18 on demand. APM costs $31 and $36. If you buy APM on the infrastructure pricing model, Datadog’s pricing FAQ says every APM host also needs Pro or Enterprise infrastructure. Ten hosts with both come to $460 a month on an annual plan, or $540 on demand, before a single log line or browser session.
Then how hosts are counted. Datadog meters hosts every hour and bills “the maximum count (high-water mark) of the lower 99% of those hourly readings.” A 30-day month has 720 hours, so only the top 7.2 hours get dropped. A scale-up that lasts a day sets your host count for the whole month. Datadog does offer a hybrid plan for autoscaling fleets, with a monthly commitment and hourly billing above it.
Containers are where it bites. “If you choose to install the Agent directly in each container, each container is counted as a host from a billing perspective,” says the billing overview. With one Agent per host, Pro includes 5 containers per host and Enterprise 10, averaged across your account. Past that, the price list charges $0.002 per container hour. Ten Pro hosts running 80 containers all month pay for 30 extra containers over 720 hours, which is $43.20.
Custom metrics count every combination of metric name and tag values, host included. In Datadog’s own example, one latency metric from two hosts, tagged by endpoint and status, is four custom metrics. Pro includes 100 per host and Enterprise 200. The price list charges $5 per 100 after that.
On top of hosts sit the usage meters. Logs cost $0.10 per GB to ingest and $1.70 per million indexed events at 15-day retention. RUM costs $0.15 per 1,000 sessions sent and $3 per 1,000 sessions kept. Error Tracking costs $25 for the first 50,000 errors.
And on-demand costs more. Those first 50,000 errors cost $36 on demand, 44% more. Indexed logs above your commitment are “charged with a 50% premium,” according to the log billing docs. A credit card signup lands on a monthly plan, and moving to annual goes through your Customer Success Manager. On demand, APM span allowances are metered by the hour, so going over in one hour “will increase billing regardless of span use during other times.”
None of this is hidden. It’s all in Datadog’s docs. It’s just a lot of meters to watch, and every new service, container or tag can move one of them.
You’d also give some things up. Errors collected inside APM traces and RUM sessions cost nothing extra in Error Tracking, and one vendor covers infrastructure, APM, logs, RUM and security.
Datadog alternatives at a glance
| Tool | Replaces | Pricing model | Best for |
|---|---|---|---|
| Grafana Cloud | Most of Datadog | Platform fee + usage | Prometheus users |
| New Relic | Most of Datadog | Per GB + per user | Big fleets, small teams |
| Dynatrace | Most of Datadog | Annual commitment | One contract, unlimited seats |
| Elastic | Most of Datadog | Per GB ingested + retained | Existing Elasticsearch users |
| Honeycomb | APM, logs and metrics | Per event | Debugging distributed requests |
| SigNoz | Most of Datadog | Per GB, $49 minimum | No host or user pricing |
| ClickStack | Logs, traces, metrics, replay | Self-host or per GB | ClickHouse and SQL users |
| Sentry | Errors, replay, tracing | Plan + per event | Web and mobile errors |
| PostHog | Errors, replay, analytics | Per unit, no base fee | Product analytics teams |
| Interfere | Web app errors and replay | Not published | Errors ranked by user impact |
Free tiers, trials and full prices are in each tool’s section below.
Full observability platforms
These are Datadog’s direct competitors. They replace most of it: infrastructure, APM, logs and, for most, RUM. Each one still counts something. What changes is what.
Grafana Cloud
Grafana Cloud bills each signal separately. The Pro plan has a $19 monthly platform fee, then usage. Metrics start at $6.50 per 1,000 active series. Logs and traces start at $0.05 per GB to process, $0.40 per GB to write and $0.10 per GB to retain. Application Observability starts at $0.025 per host hour, about $18 a host, and Frontend Observability, Grafana’s RUM, at $0.75 per 1,000 sessions after 50,000 free.
The free tier is real: 10,000 metric series, 50 GB of logs and 50 GB of traces a month, with 14 days of retention. Pro keeps metrics for 13 months and the rest for 30 days.
The part we like is the exit. Grafana, Loki, Tempo and Mimir are all on GitHub under AGPLv3, so you can run the same stack yourself later.
New Relic
New Relic answers the host question directly in its pricing FAQ: “Do you charge per host? No.” You pay for data and for people. The first 100 GB a month is free, then $0.40 per GB, or $0.60 with Data Plus.
People are the expensive part. Basic users are free and core users cost $49. Full platform users cost $10 for the first and $99 for each additional on Standard, which caps at five. Pro costs $349 a user on an annual commitment, so 20 engineers with full access cost $6,980 a month before data. A compute-based option drops user fees.
It suits a small team watching a large fleet. The free tier includes one full platform user. On the free tier, data ingest stops when you pass 100 GB until the next month or until you upgrade.
Dynatrace
Dynatrace sells one annual commitment and draws it down at rate-card prices. Full-Stack Monitoring is $58 a month per 8 GiB host, billed at $0.01 per memory-GiB-hour, so a 32 GiB host costs four times as much. Infrastructure Monitoring is $29 a host. Logs cost $0.20 per GiB to ingest, plus retention and queries. RUM is $2.25 per 1,000 sessions, or $4.50 with Session Replay.
Two things stand out. Seats are unlimited. And Dynatrace “never charges penalty-style overages.” Usage above your commitment is billed monthly at the same rates.
It fits large companies that want one contract for everything. There’s a 15-day free trial.
Elastic
Elastic Observability Serverless charges for what you ingest and what you keep. On the Complete tier, logs and traces start at $0.09 per GB ingested and $0.019 per GB a month retained. The cheaper Logs Essentials tier starts at $0.07 and $0.017. Egress is free up to 50 GB a month, then $0.05 per GB. Support above the Standard level adds 5% to 15% of your bill.
Most of Elasticsearch’s source is on GitHub under your choice of AGPLv3, SSPL or the Elastic License 2.0. If you already run it for logs or search, this is the shortest move. There’s a free trial.
Honeycomb
Honeycomb charges per event, with unlimited seats and queries. The free plan covers 20 million events a month. Pro starts at $150 a month for 50 million events and goes up to 750 million. Enterprise plans start at 10 billion events a year.
It’s built for asking new questions of traces with “unlimited custom fields,” and it includes an MCP server. We’d treat it as a replacement for Datadog APM more than for Datadog’s infrastructure dashboards.
Open source Datadog alternatives
These you can run on your own servers for free, and pay for in hardware and time. Each also sells a hosted version.
SigNoz
SigNoz is the closest open source match to Datadog’s shape: logs, traces, metrics, infrastructure and Kubernetes in one product. Its pricing page leads with “No user-based pricing,” “No host-based pricing” and “No special pricing for custom metrics.”
SigNoz Cloud starts at $49 a month, which includes $49 of usage. Logs and traces cost $0.30 per GB with 15-day retention, and metrics $0.10 per million samples with one month. Data centers are in the US, the EU and India. Enterprise starts at $4,000 a month, including a bring-your-own-cloud option.
The repository is MIT licensed outside its ee/ directory, and SigNoz has a Datadog dashboard migration tool.
ClickStack
ClickStack is ClickHouse’s observability stack: a UI, ClickHouse for storage and a preconfigured OpenTelemetry Collector. It keeps logs, traces, metrics and session replays in one place, with SQL and full-text search over high-cardinality data. The repository is MIT licensed.
The fully managed version, ClickHouse Observability, is in beta. It costs $0.10 per uncompressed GB for logs, traces and session replay, and $0.05 per million metric data points, with 30 days of retention and no monthly minimum. Those are half the list prices, which apply at general availability.
The Grafana stack belongs here too, if you’d rather run Grafana, Loki, Tempo and Mimir yourself.
If you used Datadog for errors and RUM
If you mostly pay Datadog for Error Tracking, RUM and Session Replay, you don’t need another full platform.
Sentry
Sentry covers errors, tracing, logs, profiling, and cron and uptime checks. The free Developer plan covers one user and 5,000 errors a month. Team costs $29 a month, or $26 billed annually, and includes 50,000 errors, 5 million spans, 50 replays and 5 GB of logs. You pay per event after that. You can self-host it under the FSL license.
It’s the safe pick if you ship mobile apps too. Sentry vs Datadog compares the two in detail, with a worked monthly bill.
PostHog
PostHog puts errors next to product analytics, feature flags and session replay. The first 100,000 exceptions a month are free, then $0.000370 each, falling to $0.000115 above 10 million. The first 5,000 recordings are free, then $0.0050 each. There’s no base fee and no per-seat charge, and you can set a billing limit.
PostHog is unusually honest about its gaps. Its pricing page lists advanced error grouping and mobile as reasons to pick a competitor: “Even our team thinks Sentry is better if you need mobile support.” It’s MIT licensed outside ee/, and self-hosting is officially unsupported.
Before you switch
Instrument with OpenTelemetry first. Datadog takes OTLP through its Agent, an OpenTelemetry Collector or a direct intake endpoint, and Grafana Cloud, Dynatrace, Elastic, Honeycomb and ClickStack all support OpenTelemetry. Once your services speak OTLP, changing vendors is mostly a config change, and a Collector can send to both while you compare.
One warning from Datadog’s compatibility docs: in most cases, don’t run a Datadog SDK and an OpenTelemetry SDK in the same application, “as this leads to undefined behavior.”
Where Interfere fits
Interfere is an error tracker for web apps that replaces the error tracking and session replay side of Datadog, not its infrastructure monitoring or APM. It ranks each problem by how many users and which customer accounts it affects, investigates the likely cause, and hands it to your coding agent over MCP.
More teams run like software factories now. Coding agents open pull requests all day, and shipping stops being the hard part. Knowing what broke, who it hit and whether the fix held is.
Interfere detects new problems and regressions from production errors as they arrive, sorts them by affected users and shows which companies are hit. It pulls in the traces, logs and session replays around each one and writes up the likely cause, tied to the release and code. Your agent picks it up and writes the fix. Interfere tracks recovery and reopens the problem if it comes back.
If you merge more pull requests than anyone can read line by line, that loop is how you keep your sanity.
It’s narrower than anything above. SDKs cover Next.js, React with Vite and TanStack Start, and everything else can send OpenTelemetry over OTLP/HTTP. There are no mobile SDKs, profiling, cron monitoring, uptime checks or self-hosting, and regional data residency isn’t available yet. Pricing and a free tier aren’t published. Sentry alternatives covers more error trackers, or you can try Interfere.
Which one to pick
- The whole Datadog footprint without host or user pricing: SigNoz, hosted or self-hosted. Grafana Cloud if you’d rather pay per signal.
- A cheaper alternative to Datadog for a big fleet and a small team: New Relic. No host counting, and the first 100 GB a month is free.
- One contract for a large company: Dynatrace, with unlimited seats and no overage premium.
- You already run Elasticsearch: Elastic.
- Hard questions about distributed requests: Honeycomb.
- An open source Datadog alternative you control: SigNoz, the Grafana stack or ClickStack.
- You only used Datadog for errors and replays: Sentry, especially with mobile apps, or PostHog if your product team already lives there.
- Prioritize web-app errors by real user impact and see which customers are affected: Interfere.
- You rely on Datadog APM and RUM together: stay. Errors from traces and RUM sessions cost nothing extra there.
Most of the savings come from knowing which part of Datadog you actually use. Replace that part, and leave the rest alone.




